Run Google Cloud services on-premises... what could go wrong?
Urs Hölzle, SVP of engineering, took a turn on stage to talk up a significant GCP launch called Cloud Services Platform. It's a set of services that can be run on GCP or on-premises with consistent development, management and control frameworks.
Cloud Services Platform consists of Google Kubernetes Engine (managed Kubernetes on GCP), GKE On-Prem (managed Kubernetes in your own data center, coming soon as an alpha release), and Managed Istio (an open platform for connecting, managing, and secure microservices via load balancing, policy, metrics and authentication).
It's supported by Stackdriver Monitoring and GCP Marketplace, for third party integrations. And it works with developer services like Knative (container-based serverless app framework for Kubernetes) and Cloud Build (managed CI/CD platform). Istio, having reached 1.0 in open source, now has an API management service, Apigee API Management for Istio.
Hölzle also announced GKE Policy Management, which helps admins configure and control Kubernetes, and Stackdriver Service Monitoring, which provides a real-time service graph of computing environments, monitoring aimed at service level objectives and a service dashboard to speed problem resolution and debugging. He also spoke to our sister site, The Next Platform, ahead of today's event to discuss Kubernetes, cloud, and more...
Prabhakar Raghavan, VP of engineering for G Suite, came on stage with word of enhancements to Google's productivity and collaboration suite. The most notable is the security center investigation tool, which allows G Suite admins to look into and remediate security issues. For example, Raghavan explained, it "lets an admin investigate an unusually large amount of file transfers outside a domain."
Where in the world did that come from?
Another improvement for G Suite admins is the arrival of data regions, which allows admins to specify where data resides geographically, something that matters for corporate policy and compliance.
Raghavan also announced a handful of other enhancements: Smart Compose for Gmail, introduced in May for the consumer market, will be available to Google's business customers in a few weeks; Google Docs will offer grammar suggestions to early adopters; Smart Reply has been implemented in Hangouts Chat; and Voice Commands are coming to those with Hangouts Meet Hardware – think Google Assistant commands directed at conferencing hardware.
Fei-Fei Li, chief scientist of AI and machine learning, then offered an update on everyone's favorite topic, AI and machine learning – seriously, everyone gushed about AI, with Greene calling it Google's number one opportunity.
Where most cloud services represent products without baggage, AI brings with it both hype and concern. "Even though it's a nascent science, AI is transforming industries all over the world," Li insisted.
License to kill
That's been Google's party line since it first started extolvling AI-oriented software and services a few year ago. But now that employee resignations and protests have convinced Google to distance itself from Project Maven – a Pentagon program to use AI on video images for military applications, such as remote-controlled killer drones – the web giant has become more cautious in its celebration of AI.
That caution could be heard in Li's observation that "AI is empowerment and it's our goal to democratize it." And in her remark, "We're creating technology that's not just powerful but trustworthy."
Critics of artificial intelligence argue it's just the opposite, that AI through its algorithmic opaqueness disempowers human oversight and allows bias to exist unchecked. For software to be trustworthy, it must be open to scrutiny and most AI systems aren't subject to public auditing or testing.
Consider another major AI booster, Facebook, and the effect its inscrutable automated systems have had on democracy.
But that discussion exists at a different level than the product announcements made by Li. For GCP's Cloud AutoML service, which provides developers with tools for building machine learning models, she announced that Cloud AutoML Vision, Natural Language, and Translation services have all reached beta status.
Dialogflow Enterprise Edition, for building conversation interfaces, has also moved into beta. And Google is launching a service called Contact Center AI as an alpha offering. Contact Center AI combines Dialogflow Enterprise Edition with partner integrations to deliver a more fully featured contact center automation.
Google is indeed creating technology that's powerful. We're told it's trustworthy or at least headed in that direction. Take the teleprompter's word for it. And if you'd like to speak to a robot operator, please continue to hold. Your call is very important to us. ®
PS: Cisco and Google also announced a partnership, which basically ties Cisco stuff like WebEx into Google Calendars, and so on.